Structural Basis for the Mechanism of ATP-Dependent Acetone Carboxylation Received: 6 February 2017 Florence Mus1, Brian J
Total Page：16
File Type：pdf, Size：1020Kb
www.nature.com/scientificreports OPEN Structural Basis for the Mechanism of ATP-Dependent Acetone Carboxylation Received: 6 February 2017 Florence Mus1, Brian J. Eilers2, Alexander B. Alleman1, Burak V. Kabasakal3, Jennifer N. Wells3, Accepted: 21 June 2017 James W. Murray3, Boguslaw P. Nocek4, Jennifer L. DuBois2 & John W. Peters1 Published: xx xx xxxx Microorganisms use carboxylase enzymes to form new carbon-carbon bonds by introducing carbon − dioxide gas (CO2) or its hydrated form, bicarbonate (HCO3 ), into target molecules. Acetone − carboxylases (ACs) catalyze the conversion of substrates acetone and HCO3 to form the product acetoacetate. Many bicarbonate-incorporating carboxylases rely on the organic cofactor biotin for the activation of bicarbonate. ACs contain metal ions but not organic cofactors, and use ATP to activate substrates through phosphorylation. How the enzyme coordinates these phosphorylation events and new C-C bond formation in the absence of biotin has remained a mystery since these enzymes were discovered. The first structural rationale for acetone carboxylation is presented here, focusing on the 360 kDa (αβγ)2 heterohexameric AC from Xanthobacter autotrophicus in the ligand-free, AMP-bound, and acetate coordinated states. These structures suggest successive steps in a catalytic cycle revealing that AC undergoes large conformational changes coupled to substrate activation by ATP to perform C-C bond ligation at a distant Mn center. These results illustrate a new chemical strategy for the conversion of CO2 into biomass, a process of great significance to the global carbon cycle. Carboxylases are enzymes that catalyze the incorporation of CO2 into organic substrates. Assimilatory carboxy- lases use the carboxylation reaction to directly transform diverse carbon sources into central metabolites as part of autotrophic (photosynthetic) or heterotrophic metabolic pathways. The latter pathways are essential for the biological assimilation of often chemically intransigent, poorly activated organic compounds1. Most enzymatic carboxylation reactions follow the same mechanistic principle: nucleophilic activation of substrates and electro- 2 philic activation of CO2 . However, the stepwise mechanisms of carboxylation reactions differ in essential ways with respect to co-substrate, co-factor or metal requirements. Knowledge of these mechanisms provides the basis for an increased fundamental understanding of carboxylation chemistry, and contributes to future strategies for 3 CO2 capture. These in turn may mitigate the effects of increasing concentrations of CO2 on the global climate . Acetone carboxylases are assimilatory carboxylases that catalyze the conversion of substrates acetone and − HCO3 to form the product acetoacetate (Fig. 1a) allowing bacteria to incorporate this small, volatile and envi- ronmentally toxic ketone into biomass. In general, bicarbonate-dependent carboxylases catalyze the net dehy- 4 dration of H2CO3, retaining CO2 as a biotin adduct . However, ACs purified from multiple bacterial sources have been shown to be free of biotin or any other organic cofactor, instead containing quantities of manganese, zinc, and iron within a heteromultimeric protein complex5–8. These carboxylases were also shown to convert ATP to AMP and two inorganic phosphate anions, suggesting that they catalyze the phosphorylation-dependent activa- tion of both carbon substrates from a single nucleotide9. This reaction sets ACs apart from the phylogenetically related acetophenone carboxylases (APCs). APC hydrolyses two ATP to ADP in order to activate acetophenone and bicarbonate10. The AC β subunit and APC α subunits share homology and both possess nucleotide-binding sites. The structure of APC was recently deter- mined, revealing that the two MgATP binding sites that are the proposed sites for the activation of acetophenone and bicarbonate are separated by ~50 Å. A large conformational shift was proposed to bring the two phosphoryl- ated intermediates in closer proximity for catalysis11. 1Insitutite of Biological Chemistry, Washington State University, Pullman, WA, 99164, USA. 2Department of Chemistry and Biochemistry, Montana State University, Bozeman, MT, 59717, USA. 3Department of Life Sciences, Imperial College, London, SW7 2AZ, UK. 4Structural Biology Center, Argonne National Laboratory, Argonne, IL, 60439, USA. Florence Mus and Brian J. Eilers contributed equally to this work. Correspondence and requests for materials should be addressed to J.W.P. (email: [email protected]) SCIENTIFIC REPORTS | 7: 7234 | DOI:10.1038/s41598-017-06973-8 1 www.nature.com/scientificreports/ Figure 1. Overall reaction scheme and crystal structure of AMP bound AC. (a) Reaction of ACs. The sequential phosphorylation of the products acetone and then bicarbonate by the γ then β phosphates from ATP, respectively, creates the highly reactive intermediates phosphoenolacetone and carboxyphosphate. These are proposed to react together at the Mn2+ active site to create acetoacetate and two molecules of inorganic phosphate (Pi). (b) The overall structure of the (αβγ)2 heterohexameric AC enzyme is shown: α subunits (green), β subunits (blue), γ subunits (violet). One monomer is transparent to indicate the dimer interface as well as the nucleotide binding site. Mn2+, Zn2+ and AMP binding sites are indicated with arrows. Distinct from APC, AC from Xanthobacter autotrophicus Py2 has been shown to activate both acetone and bicarbonate with a single ATP, presumably using both γ and β phosphates in a sequential fashion, for a ligation reaction at a Mn-containing catalytic site9. The lack of structural information on ACs has been a significant impediment in formulating plausible mechanistic hypotheses. Here we present the first x-ray crystal struc- tures of the 360 kDa (αβγ)2 heterohexameric AC. Surprisingly, the ATP binding site and the catalytically essen- tial Mn cofactor, long assumed to be adjacent to one another on the basis of previous spectroscopic studies12, are separated by ~40 Å. A series of structures in the ligand free, AMP-bound, and acetate-coordinated states, which we represent as approximate mechanistically relevant states, allows the inference of a new mechanism for enzyme-mediated CO2 capture and functionalization. Results and Discussion Like many carboxylases, AC is a heteromultimeric enzyme complex. Its architecture consists of two heterotri- meric αβγ subunits joined by the interacting α-subunits to form a dimeric core (Figs 1b and S1). The α subunit (75 kDa) shares a large interface with the β subunit (85 kDa). The γ subunit (20 kDa) interacts mostly with the α subunit and shares a small contact with the β subunit through a helix at the carboxyl end of the γ subunit. This interaction area between all three subunits creates a cleft on the solvent surface. AC’s α subunit shows high structural similarity to APCβ and contains similar internal folding domains, including the α/β interface. This interface is anchored by the large helix-2 on the α-subunit that leads into the polyproline II-like helix bundle similar to what is observed in APC11. The β subunit interface is made up of two α/β sandwich-like domains with exposed α-helices that interact directly with helix-2 from the α subunit. The nucleotide-binding β subunit has low structural similarity with kinases such as glycerol kinase (PDB: 3FLC, DALI-Z score 10.9), pantothenate kinase (PDB: 3BF3, DALI-Z score 12.2), and L-rhamnulose kinase (PDB: 2CGL, DALI-Z score 9.1), which share similar nucleotide-binding pockets13–15. The β subunit also shares nucleotide binding residues with mutually homologous subunits APCα and APCα’ (PDB: 5L9W, DALI-Z score 36.5 and 46.4 respectively). The γ-subunits, which bear homology to nucleotide binding Yippee-like domains, contain conserved cysteine residues (Cys74, Cys76, Cys124, Cys127) that form a 4-coordinate Zn binding site16. The role of the γ-subunit is not clear, and the Zn ion is not predicted, from prior data, to have a catalytic role. X. autotrophicus Py2 AC (XaAC) differs from that reported for Aromatoleum aromaticum (AaAC). XaAC has been reported as having 1.3 mol Mn and 0.7 mol Fe per mol of enzyme and AaAC has been reported to have significantly different metal content with no Mn and 2.2 mol Fe per mol of enzyme8. In addition, the stoichiometry for ATP consumption reported is also different with XaAC requiring only 1 ATP for acetoacetate formation and AaAC requiring 2 ATP5, 8. These differences are surprising given the high level of sequence conservation of the two enzymes. However, the observed differences in metal content in ACs purified from different sources are likely related to the differences in stoichiometry for ATP. Although the presence of Fe has been observed in previous studies of X. autotrophicus Py2 AC (0.7 Fe/ 5 (αβγ)2) , K-edge anomalous difference data were not consistent with Fe in the structures presented here (see Supplementary Data, Table S2). The lack of Fe in the crystal data and previous studies by Boyd et al.12 suggest that Fe may associate with the enzyme in the absence of manganese but is not a catalytically effective subsitute12. Both the recombinant and native X. autotrophicus Py2 acetone carboxylases used for crystallization studies actively SCIENTIFIC REPORTS | 7: 7234 | DOI:10.1038/s41598-017-06973-8 2 www.nature.com/scientificreports/ Figure 2. Conformational shift upon nucleotide binding. (a) Ligand-free structure (α subunits: olive; β subunits: violet; γ subunits: limon) showing a substrate channel (grey)